Update src/streamlit_app.py
Browse files- src/streamlit_app.py +22 -98
src/streamlit_app.py
CHANGED
|
@@ -7,20 +7,6 @@ import streamlit as st
|
|
| 7 |
import torch
|
| 8 |
from document_registry import DocumentRegistry
|
| 9 |
|
| 10 |
-
# --- Initialize session keys to avoid rerun issues ---
|
| 11 |
-
if "registry" not in st.session_state:
|
| 12 |
-
st.session_state.registry = DocumentRegistry()
|
| 13 |
-
if "active_doc" not in st.session_state:
|
| 14 |
-
st.session_state.active_doc = None
|
| 15 |
-
if "chunks" not in st.session_state:
|
| 16 |
-
st.session_state.chunks = []
|
| 17 |
-
if "embeddings" not in st.session_state:
|
| 18 |
-
st.session_state.embeddings = None
|
| 19 |
-
if "index" not in st.session_state:
|
| 20 |
-
st.session_state.index = None
|
| 21 |
-
if "query_suggestions_fixed" not in st.session_state:
|
| 22 |
-
st.session_state.query_suggestions_fixed = []
|
| 23 |
-
|
| 24 |
|
| 25 |
# ==========================================================
|
| 26 |
# β
PAGE CONFIGS
|
|
@@ -100,44 +86,6 @@ Output: Write each question on a new line. Do not invent facts β base question
|
|
| 100 |
except Exception:
|
| 101 |
return ["How do I start using this guide?", "What does this document cover?"]
|
| 102 |
|
| 103 |
-
# --- Function to activate a document from the registry ---
|
| 104 |
-
def activate_document(doc_name):
|
| 105 |
-
"""Load an existing registered document into the active context."""
|
| 106 |
-
registry = st.session_state.registry
|
| 107 |
-
doc = registry.get_doc(doc_name)
|
| 108 |
-
|
| 109 |
-
if not doc:
|
| 110 |
-
st.warning(f"Document {doc_name} not found in registry.")
|
| 111 |
-
return
|
| 112 |
-
|
| 113 |
-
# Load all available preprocessed data
|
| 114 |
-
st.session_state.active_doc = doc
|
| 115 |
-
st.session_state.chunks = doc.get("chunks", [])
|
| 116 |
-
st.session_state.embeddings = doc.get("embeddings")
|
| 117 |
-
st.session_state.index = doc.get("index")
|
| 118 |
-
|
| 119 |
-
# --- Handle TOC safely ---
|
| 120 |
-
# Some docs may not have toc saved in registry; fallback to empty list
|
| 121 |
-
toc_data = doc.get("toc", [])
|
| 122 |
-
if not toc_data:
|
| 123 |
-
st.info(f"β οΈ No TOC found for {doc_name}. Using empty TOC for suggestions.")
|
| 124 |
-
|
| 125 |
-
# --- Regenerate suggestions safely ---
|
| 126 |
-
try:
|
| 127 |
-
st.session_state.query_suggestions_fixed = generate_dynamic_suggestions_from_toc(
|
| 128 |
-
toc_data,
|
| 129 |
-
st.session_state.chunks,
|
| 130 |
-
doc.get("name", "Unknown Document")
|
| 131 |
-
)
|
| 132 |
-
except Exception as e:
|
| 133 |
-
# Defensive guard: even if suggestion generator fails, app won't break
|
| 134 |
-
st.warning(f"Suggestion generation failed: {e}")
|
| 135 |
-
st.session_state.query_suggestions_fixed = []
|
| 136 |
-
|
| 137 |
-
# --- Trigger a clean UI rerun so the switch shows immediately ---
|
| 138 |
-
st.rerun()
|
| 139 |
-
|
| 140 |
-
|
| 141 |
|
| 142 |
# ==========================================================
|
| 143 |
# π¨ STYLING β REVERT TO ORIGINAL
|
|
@@ -189,6 +137,7 @@ h1, h2, h3 {color: #f3f4f6; font-weight: 600;}
|
|
| 189 |
# π§ SIDEBAR
|
| 190 |
# ==========================================================
|
| 191 |
with st.sidebar:
|
|
|
|
| 192 |
st.markdown("### π§ Response Style")
|
| 193 |
mode = st.radio(
|
| 194 |
"",
|
|
@@ -199,50 +148,23 @@ with st.sidebar:
|
|
| 199 |
|
| 200 |
st.markdown("---")
|
| 201 |
|
| 202 |
-
#
|
| 203 |
-
|
| 204 |
-
|
| 205 |
-
|
| 206 |
-
|
| 207 |
-
|
| 208 |
-
|
| 209 |
-
|
| 210 |
-
|
| 211 |
-
|
| 212 |
-
|
| 213 |
-
chunks = doc.get("num_chunks", "?")
|
| 214 |
-
toc_source = doc.get("toc_source", "β")
|
| 215 |
-
|
| 216 |
-
# Use an expander for each document
|
| 217 |
-
with st.expander(f"{i}. {doc_name}", expanded=False):
|
| 218 |
-
st.markdown(f"- π§© **Chunks:** {chunks}")
|
| 219 |
-
st.markdown(f"- ποΈ **TOC Source:** {toc_source}")
|
| 220 |
-
|
| 221 |
-
# If this document is currently active, show status
|
| 222 |
-
if (
|
| 223 |
-
"active_doc" in st.session_state
|
| 224 |
-
and st.session_state.active_doc
|
| 225 |
-
and isinstance(st.session_state.active_doc, dict)
|
| 226 |
-
and st.session_state.active_doc.get("name") == doc_name
|
| 227 |
-
|
| 228 |
-
):
|
| 229 |
-
st.success("β
Active Document")
|
| 230 |
-
else:
|
| 231 |
-
# Button to activate this document
|
| 232 |
-
st.button(
|
| 233 |
-
"Activate",
|
| 234 |
-
key=f"activate_{doc_name}",
|
| 235 |
-
on_click=activate_document,
|
| 236 |
-
args=(doc_name,),
|
| 237 |
-
)
|
| 238 |
-
else:
|
| 239 |
-
st.caption("π No documents registered yet.")
|
| 240 |
else:
|
| 241 |
-
st.caption("π No
|
| 242 |
|
| 243 |
st.markdown("---")
|
| 244 |
|
| 245 |
-
#
|
| 246 |
show_dev = st.checkbox("Show advanced settings (for developers)", value=False)
|
| 247 |
if show_dev:
|
| 248 |
st.markdown("### βοΈ Developer Options")
|
|
@@ -255,7 +177,7 @@ with st.sidebar:
|
|
| 255 |
st.markdown("---")
|
| 256 |
st.caption("β¨ Built by Shubham Sharma")
|
| 257 |
|
| 258 |
-
#
|
| 259 |
if show_dev:
|
| 260 |
st.markdown("---")
|
| 261 |
with st.expander("π§© Developer Insights", expanded=False):
|
|
@@ -278,6 +200,7 @@ with st.sidebar:
|
|
| 278 |
st.caption(f"{len(st.session_state.get('chunks', []))} chunks processed.")
|
| 279 |
|
| 280 |
|
|
|
|
| 281 |
# ==========================================================
|
| 282 |
# π§ SESSION STATE SAFETY INITIALIZATION
|
| 283 |
# ==========================================================
|
|
@@ -351,12 +274,13 @@ else:
|
|
| 351 |
embeddings = cache_embeddings(doc_name, chunks, embed_chunks)
|
| 352 |
index = build_faiss_index(embeddings)
|
| 353 |
|
| 354 |
-
|
| 355 |
-
|
|
|
|
|
|
|
| 356 |
|
| 357 |
-
registry
|
| 358 |
-
|
| 359 |
-
st.session_state["active_doc"] = doc_id
|
| 360 |
|
| 361 |
status.success("β
Document processed successfully β all set to query your assistant!")
|
| 362 |
|
|
|
|
| 7 |
import torch
|
| 8 |
from document_registry import DocumentRegistry
|
| 9 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
|
| 11 |
# ==========================================================
|
| 12 |
# β
PAGE CONFIGS
|
|
|
|
| 86 |
except Exception:
|
| 87 |
return ["How do I start using this guide?", "What does this document cover?"]
|
| 88 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 89 |
|
| 90 |
# ==========================================================
|
| 91 |
# π¨ STYLING β REVERT TO ORIGINAL
|
|
|
|
| 137 |
# π§ SIDEBAR
|
| 138 |
# ==========================================================
|
| 139 |
with st.sidebar:
|
| 140 |
+
# --- Response Style ---
|
| 141 |
st.markdown("### π§ Response Style")
|
| 142 |
mode = st.radio(
|
| 143 |
"",
|
|
|
|
| 148 |
|
| 149 |
st.markdown("---")
|
| 150 |
|
| 151 |
+
# --- Registered Documents (simple, user-friendly) ---
|
| 152 |
+
registry = st.session_state.get("registry", None)
|
| 153 |
+
registered_docs = registry.list_docs() if registry and hasattr(registry, "list_docs") else []
|
| 154 |
+
|
| 155 |
+
if registered_docs:
|
| 156 |
+
st.markdown("### π Uploaded Documents")
|
| 157 |
+
for i, doc in enumerate(registered_docs, start=1):
|
| 158 |
+
doc_name = doc.get("name", f"Document {i}")
|
| 159 |
+
chunks = doc.get("num_chunks", "?")
|
| 160 |
+
toc_source = doc.get("toc_source", "β")
|
| 161 |
+
st.markdown(f"**{i}. {doc_name}** β {chunks} chunks *(TOC: {toc_source})*")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 162 |
else:
|
| 163 |
+
st.caption("π No documents uploaded yet.")
|
| 164 |
|
| 165 |
st.markdown("---")
|
| 166 |
|
| 167 |
+
# --- Developer Options ---
|
| 168 |
show_dev = st.checkbox("Show advanced settings (for developers)", value=False)
|
| 169 |
if show_dev:
|
| 170 |
st.markdown("### βοΈ Developer Options")
|
|
|
|
| 177 |
st.markdown("---")
|
| 178 |
st.caption("β¨ Built by Shubham Sharma")
|
| 179 |
|
| 180 |
+
# --- Developer Insights (optional, hidden by default) ---
|
| 181 |
if show_dev:
|
| 182 |
st.markdown("---")
|
| 183 |
with st.expander("π§© Developer Insights", expanded=False):
|
|
|
|
| 200 |
st.caption(f"{len(st.session_state.get('chunks', []))} chunks processed.")
|
| 201 |
|
| 202 |
|
| 203 |
+
|
| 204 |
# ==========================================================
|
| 205 |
# π§ SESSION STATE SAFETY INITIALIZATION
|
| 206 |
# ==========================================================
|
|
|
|
| 274 |
embeddings = cache_embeddings(doc_name, chunks, embed_chunks)
|
| 275 |
index = build_faiss_index(embeddings)
|
| 276 |
|
| 277 |
+
registry = st.session_state.get("registry")
|
| 278 |
+
if not registry:
|
| 279 |
+
registry = DocumentRegistry()
|
| 280 |
+
st.session_state["registry"] = registry
|
| 281 |
|
| 282 |
+
registry.register(temp_path, chunks, embeddings, index)
|
| 283 |
+
|
|
|
|
| 284 |
|
| 285 |
status.success("β
Document processed successfully β all set to query your assistant!")
|
| 286 |
|